Data Science Challenge 2023 - Cardiac Electrophysiology

This repository contains the notebooks and code for the 2023 Data Science Challenge (DSC) at Lawrence Livermore National Laboratory (LLNL).

Link : https://data-science.llnl.gov/dsc
Authors :
- Mikel Landajuela (landajuelala1@llnl.gov)
- Cindy Gonzales (gonzales72@llnl.gov)

Description

Training

To train the model, use the following command

python train.py --config_path=configs/config.yaml --save_dir=<save_dir> --tag=<tag>

If you do not prefer your own <save_dir>, the program will create one for you in the /tmp/ folder.

If you'd rather submit a slurm script, run the following command

sbatch scripts/sbatch_train.sh -c configs/config.yaml -d <save_dir> -t TAG

Just make sure, when you are using the slurm batch script, to run it on the right group with the correct access-rights.

Evaluation

To evaluate the model, use the following command

    python eval.py --save_dir=<SAVE_DIR>

You do not have to specify the configs.yaml file. The SAVE_DIR folder already has one. But you may, in case the configs file is different from the one in the SAVE_DIR folder.

How to add your own model+dataset ?

To add your own model, you would need to have a dataset file and a model file.

Once you have the dataset file, place it under <dataset> folder.

You may keep the model file under <models> folder.

For the model file, make sure the forward is replaced with _forward and nn.Module is replaced with BaseModel. Add the line (from base import BaseModel) to the preamble.

Then, in builder.py, you need to add the model

from datasets import <File-Name>
from models import <Model-FileName>
...
DATASETS ={
    ...
    'yourdata': <File-Name>.<Class-Name>
    ...
}

from models import ..., YourModelFileName
MODELS = {
    ...
    'yourmodel':<Model-FileName>.<Class-Name>
}

to the MODEL dictionary, and add the dataset to the DATASETS dictionary.

Contact